CN111624494B - Battery analysis method and system based on electrochemical parameters - Google Patents
Battery analysis method and system based on electrochemical parameters Download PDFInfo
- Publication number
- CN111624494B CN111624494B CN202010314133.1A CN202010314133A CN111624494B CN 111624494 B CN111624494 B CN 111624494B CN 202010314133 A CN202010314133 A CN 202010314133A CN 111624494 B CN111624494 B CN 111624494B
- Authority
- CN
- China
- Prior art keywords
- parameters
- battery
- data
- macroscopic
- electrochemical
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/367—Software therefor, e.g. for battery testing using modelling or look-up tables
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/371—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC] with remote indication, e.g. on external chargers
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/382—Arrangements for monitoring battery or accumulator variables, e.g. SoC
- G01R31/3842—Arrangements for monitoring battery or accumulator variables, e.g. SoC combining voltage and current measurements
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/389—Measuring internal impedance, internal conductance or related variables
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/392—Determining battery ageing or deterioration, e.g. state of health
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Secondary Cells (AREA)
- Tests Of Electric Status Of Batteries (AREA)
- Charge And Discharge Circuits For Batteries Or The Like (AREA)
Abstract
The invention relates to a battery analysis method and system based on electrochemical parameters. According to the method, the acquired macroscopic parameters of the battery and the electrochemical-based microscopic parameters obtained through further calculation by a curve fitting algorithm are coupled, so that the characteristic parameters input into the trained machine learning model contain more effective information, two-stage data cleaning is performed through edge calculation and cloud calculation, the prediction accuracy of the machine learning model is fundamentally improved, fault early warning and out-of-control alarming can be realized according to outlier detection of real-time data through analysis of battery big data, a more accurate prediction model can be obtained through training of mass historical data, and accurate estimation of the battery state is realized.
Description
Technical Field
The invention belongs to the technical field of battery analysis, and particularly relates to a battery analysis method and system based on electrochemical parameters.
Background
Energy crisis and environmental pollution are two major problems facing the world at present, governments and automobile manufacturers are actively taking measures to deal with the problems, renewable energy technology is developed as an important method for solving the series of problems, and as an important branch of new energy technology, the green, clean and efficient new energy automobile technology can reduce the dependence on fossil energy and reduce the emission of pollutants and greenhouse gases, and finally replaces the traditional fuel automobile. Compared with the traditional automobile, the electric automobile has the advantages of cleanness, energy saving, no pollution and low noise, and the core for developing the electric automobile lies in solving the problems of cost, endurance and safety, which are mainly limited by the performance and management and control of the power battery for the electric automobile. In other words, the technical bottlenecks faced by the power battery and the management system thereof directly restrict the large-scale popularization and application of the electric vehicle.
At present, fault detection and state prediction of a power battery mainly utilize a current, voltage and temperature sensor-based measuring method, although detection can be realized in most states, the research and characterization of the internal mechanism evolution of the battery cannot be met, and therefore high-precision prediction cannot be realized; the microstructure of the battery can be observed based on an electrochemical experimental means, but the electrochemical experimental means is too dependent on stable and controllable laboratory conditions, so that the comprehensive detection of multiple working conditions cannot be met. More importantly, the two traditional methods are generally intervened when or after a fault occurs, and real-time accurate fault early warning and state prediction cannot be guaranteed.
Disclosure of Invention
Aiming at the technical problems that real-time accurate fault early warning and state prediction cannot be guaranteed in the prior art, the invention provides a battery analysis method based on electrochemical parameters. The invention also provides a battery analysis system based on the electrochemical parameters.
The technical scheme of the invention is as follows:
a method for battery analysis based on electrochemical parameters, said method comprising the steps of:
acquiring macroscopic parameters of the battery;
calculating the battery microscopic parameters based on the electrochemical principle through a curve fitting algorithm according to the battery macroscopic parameters;
performing data pre-cleaning on the battery macroscopic parameters and the battery microscopic parameters through edge calculation to obtain pre-cleaned data, sending the pre-cleaned data to a cloud big data platform, and performing cloud data cleaning on the pre-cleaned data sent to the cloud big data platform to obtain data streams which are real values;
detecting and calculating the data streams which are all real values through an outlier detection algorithm to obtain the number of outlier data;
comparing the number of the outlier data with a preset threshold value of an alarm level, and determining whether to alarm and the alarm level according to a comparison result;
and inputting the data streams which are all real values into a trained machine learning model, comparing the fault information of the outlier with a fault set by using the machine learning model through a clustering method, and predicting the fault type according to a clustering result to realize the state estimation of the battery.
Further, the number of the outliers is compared with a preset threshold of an alarm level, whether the alarm is given and the alarm level is determined according to the comparison result, the number of the outliers corresponds to the alarm level one by one, and the alarm result is obtained by comparing the number of the outliers with the preset threshold of the corresponding alarm level in real time or within a certain time interval.
Further, a plurality of sensors at the vehicle end are used for acquiring macroscopic parameters of the battery, wherein the macroscopic parameters comprise: current, voltage, temperature, capacity and internal resistance of the battery.
Further, the microscopic parameters include: the system comprises differential thermal voltage analysis curve parameters, incremental capacity analysis curve parameters, differential voltage analysis curve parameters, charging and discharging voltage platform analysis curve parameters and parameter identification electrochemical impedance spectrum parameters.
Further, the cloud data cleaning comprises spare assignment, error value correction and logic verification.
Further, the predicting the type of the fault which may occur according to the clustering result comprises predicting a state of charge (SOC), a state of energy (SOE), a state of power (SOP) and/or a state of life (SOH) according to the clustering result.
A battery analysis system based on electrochemical parameters, the system comprising a macroscopic parameter acquisition device for acquiring macroscopic parameters of a battery, the system further comprising:
the microscopic parameter calculating device is used for calculating the microscopic parameters of the battery based on the electrochemical principle through a curve fitting algorithm according to the macroscopic parameters of the battery;
the data cleaning device is used for performing data pre-cleaning on the battery macroscopic parameters and the battery microscopic parameters through edge calculation to obtain pre-cleaned data and sending the pre-cleaned data to the cloud big data platform, and performing cloud data cleaning on the pre-cleaned data sent to the cloud big data platform to obtain data streams which are real values;
the outlier detection device is used for detecting and calculating the data streams which are all real values through an outlier detection algorithm to obtain the number of outlier data;
the real-time alarm device is used for comparing the number of the cluster data with a preset threshold value of the alarm level and determining whether to alarm and the alarm level according to the comparison result;
and the detection and prediction device is used for inputting the data streams which are all real values into a trained machine learning model, comparing the fault information of the outlier with a fault set through a clustering method, and predicting the fault type according to a clustering result so as to realize the state estimation of the battery.
Further, the macroscopic parameter acquisition device acquires macroscopic parameters of the battery through a plurality of sensors at the vehicle end, and the macroscopic parameters comprise: current, voltage, temperature, capacity and internal resistance of the battery; the microscopic parameters calculated by the microscopic parameter calculating means include: the system comprises differential thermal voltage analysis curve parameters, incremental capacity analysis curve parameters, differential voltage analysis curve parameters, charging and discharging voltage platform analysis curve parameters and parameter identification electrochemical impedance spectrum parameters.
Furthermore, the data cleaning device comprises an edge calculating device which is arranged at the output end of the vehicle-mounted power battery and used for carrying out format verification on the collected battery macro parameters and the electrochemical micro parameters subjected to calculation processing and assigning values to null values of data which are not collected due to sensor faults so as to realize data pre-cleaning.
The invention has the following technical effects:
the invention provides a battery analysis method based on electrochemical parameters, which utilizes the coupling of macroscopic parameters and microscopic parameters to ensure that a data stream input into a cloud large data platform contains more effective information, thus fundamentally improving the real-time prediction precision of a machine learning model, and after the data stream is transmitted to the cloud large data platform, the prediction model, namely the trained machine learning model can train and analyze real-time streaming data; the battery microscopic parameters based on the electrochemical principle can contain a large amount of internal mechanism information of the battery, and compared with the traditional single macroscopic parameter analysis and detection method, the battery microscopic parameters are more comprehensive and can realize more accurate prediction effect; according to the battery analysis method based on the electrochemical parameters, real-time electrochemical characterization parameters, namely battery macro parameters, can be calculated by collecting data such as current, voltage, temperature, capacity, internal resistance and the like of a battery, electrochemical micro parameters representing the internal mechanism change of the battery and the macro parameters are jointly used as input characteristic variables, and fault early warning and out-of-control alarm can be realized according to the outlier detection of the real-time data through the analysis of battery big data; a more accurate prediction model, namely a machine learning model, can be obtained through training of massive historical data, and the state of the battery is accurately estimated. The electrochemical performance, the thermodynamic performance, the mechanical fault and the like of the battery can be effectively disclosed, a more comprehensive real-time fault alarm effect is realized, the failure rate and the false alarm rate are reduced, the reliability of data can be increased by the electrochemical parameter-based battery analysis method provided by the invention through two-stage data cleaning of the edge end and the cloud end, higher-quality input is provided for a machine learning model, and the quality of a battery performance detection prediction result can be improved by utilizing the machine learning model by relying on a strong cloud end big data platform.
The invention also provides a battery analysis system based on electrochemical parameters, which corresponds to the battery analysis method based on electrochemical parameters and can be considered as a system for realizing the battery analysis method based on electrochemical parameters, the system comprises a macroscopic parameter acquisition device, a microscopic parameter calculation device for calculating battery microscopic parameters based on electrochemical principles according to the battery macroscopic parameters, a cloud data cleaning device for performing data pre-cleaning on the battery macroscopic parameters and the battery microscopic parameters to obtain pre-cleaned data and sending the pre-cleaned data to a cloud big data platform, the cloud data cleaning device for performing higher computation on the pre-cleaned data sent to the cloud big data platform to realize two-stage data cleaning to obtain real-value data streams, and the detection and calculation on the real-value data streams through an outlier detection algorithm, the system comprises an outlier detection device for obtaining the number of outlier data, a real-time alarm device for comparing the number of the outlier data with a preset threshold value of an alarm grade, a machine learning model for inputting a data stream which is real-valued, a detection prediction device for determining whether to alarm and the alarm grade according to a comparison result, a clustering method for comparing the fault information of the outlier with a fault set and predicting the fault type which is possibly generated according to the clustering result, wherein all devices of the system are mutually cooperated to work, and a microscopic parameter calculation device is used for coupling a macroscopic parameter with a microscopic parameter, so that the data stream input to a cloud large data platform contains more effective information, the real-time prediction precision of the machine learning model is fundamentally improved, and the machine learning model can train and analyze real-time streaming data after the data stream is transmitted to the cloud large data platform, the method comprises the steps that an outlier detection device gives a fault alarm in real time, a machine learning model is used for comparing fault information of an outlier with a fault set through a clustering method, the type of a fault which possibly occurs is predicted according to a clustering result, the model can also be used for predicting the battery state in a future period of time through a data stream input in real time, more accurate state prediction can be achieved, and guarantee is provided for energy management and control of the whole vehicle; the micro parameter calculation device calculates and obtains the battery micro parameters based on the electrochemical principle, and the electrochemical parameters can represent the fundamental characteristics of the internal reaction of the battery, so that the battery state can be more accurately and comprehensively depicted, the reliability is higher, compared with the traditional single macro parameter analysis and detection method, the method is more comprehensive, the more accurate prediction effect can be realized, the electrochemical performance, the thermodynamic performance, the mechanical fault and the like of the battery can be effectively disclosed, the more comprehensive real-time fault alarm effect is realized, the failure rate and the false alarm rate are reduced, the reliability of data can be increased through the data cleaning device, higher quality input is provided for a machine learning model, and the quality improvement of the battery performance detection and prediction result can be realized by utilizing the machine learning model by depending on a powerful cloud large data platform.
Drawings
FIG. 1 is a flow chart of an electrochemical parameter analysis method according to the present invention.
Fig. 2 is a schematic diagram of the device structure of a battery analysis system based on electrochemical parameters according to the present invention.
Detailed Description
For a clearer understanding of the contents of the present invention, reference will be made to the accompanying drawings and examples.
The invention provides a battery analysis method based on electrochemical parameters, as shown in figure 1, the method comprises the following steps: acquiring macroscopic parameters of the battery; calculating the battery microscopic parameters based on the electrochemical principle through a curve fitting algorithm according to the battery macroscopic parameters; performing data pre-cleaning on the macro parameters and the micro parameters of the battery through edge calculation to obtain pre-cleaned data, sending the pre-cleaned data to a cloud big data platform, and performing cloud data cleaning on the pre-cleaned data sent to the cloud big data platform through cloud calculation to obtain data streams which are real values; detecting and calculating the data streams which are all real values through an outlier detection algorithm to obtain the number of outlier data; comparing the number of the cluster data with a preset threshold value of the alarm level, and determining whether to alarm and the alarm level according to the comparison result; and inputting the data streams which are all real values into a trained machine learning model, comparing the fault information of the outlier with a fault set by using the machine learning model through a clustering method, and predicting the fault type according to a clustering result to realize the state estimation of the battery.
Specifically, in the embodiment, the cloud big data platform has the characterization capability of the internal electrochemical mechanism of the battery, and can more accurately depict the change rule of the battery in the physical world, so that the cloud big data platform has great potential in the aspects of real-time alarm, fault early warning and state prediction.
Specifically, the battery macro parameters are current, voltage, temperature, capacity and internal resistance parameters which are acquired by a macro parameter acquisition device in a vehicle-mounted BMS data terminal in real time at a certain frequency, and the parameters send a vehicle-mounted communication device to a cloud big data platform. On one hand, after the macroscopic parameters of the battery are transmitted to the cloud platform, the state of the battery can be monitored through an outlier detection algorithm (including but not limited to a detection method based on statistics, clustering and density), and a grading alarm system can be implemented for the detection condition corresponding to each type of parameters; on the other hand, the battery macroscopic parameters can calculate the related electrochemical parameters, namely the battery microscopic parameters, and are finally input into the data dimension expansion of the machine learning model.
Specifically, the microscopic parameters of the battery are parameters obtained by performing curve fitting on a DTV curve, an incremental capacity analysis ICA curve, a differential voltage analysis DV curve, a charge-discharge voltage platform analysis UU curve and variation curves of parameters such as ohmic resistance, polarization resistance and polarization capacitance obtained by parameter identification according to the macroscopic parameters of the battery and based on electrochemical differential thermal voltage.
Specifically, the alarm level can be customized according to design requirements and national standards, and is generally 3-4.
First-stage: and (5) powering off in an emergency.
And (2) second stage: requesting power down.
Third-stage: the power is limited.
And (4) fourth stage: and (6) battery early warning.
In particular, machine learning models include, but are not limited to, some regression, classification, object detection algorithms based on artificial intelligence techniques. Such as BP neural network, LSTM neural network, SVR support vector machine, and boosting, bagging ensemble learning methods or other optimized solutions based on these algorithms to improve performance.
Based on the embodiment of the invention, the electrochemical parameter-based battery analysis method provided by the invention has the advantages that by utilizing the coupling of macroscopic parameters and microscopic parameters, a data stream input into a cloud big data platform contains more effective information, so that the real-time prediction precision of a machine learning model is fundamentally improved, and after the data stream is transmitted to the cloud big data platform, the prediction model, namely the trained machine learning model can train and analyze real-time streaming data, mainly through real-time fault alarm of outlier detection, the data and fault set data can be compared through a clustering algorithm to analyze fault types, the model can also predict the battery state in the future period of time by utilizing the real-time input data stream, more accurate state prediction can be realized, and guarantee is provided for energy management and control of the whole vehicle; the battery microscopic parameters based on the electrochemical principle can contain a large amount of internal mechanism information of the battery, and compared with the traditional single macroscopic parameter analysis and detection method, the battery microscopic parameters are more comprehensive and can realize more accurate prediction effect; according to the battery analysis method based on the electrochemical parameters, real-time electrochemical characterization parameters, namely battery macro parameters, can be calculated by collecting data such as current, voltage, temperature, capacity, internal resistance and the like of a battery, electrochemical micro parameters representing the internal mechanism change of the battery and the macro parameters are jointly used as input characteristic variables, and fault early warning and out-of-control alarm can be realized according to the outlier detection of the real-time data through the analysis of battery big data; a more accurate prediction model, namely a machine learning model, can be obtained through training of massive historical data, and the state of the battery is accurately estimated. The electrochemical performance, the thermodynamic performance, the mechanical fault and the like of the battery can be effectively disclosed, a more comprehensive real-time fault alarm effect is realized, the failure rate and the false alarm rate are reduced, the reliability of data can be increased by the electrochemical parameter-based battery analysis method provided by the invention through two-stage data cleaning of the edge end and the cloud end, higher-quality input is provided for a machine learning model, and the quality of a battery performance detection prediction result can be improved by utilizing the machine learning model by relying on a strong cloud end big data platform.
In the above embodiment, the mathematical expression of the differential thermal voltage analysis DTV curve is as follows:
the DTV curve reflects the phase change of the internal electrode of the battery and the heat generation change caused by other electrochemical side reactions by using the temperature change in the constant current charging and discharging process, and the change of the voltage difference reflects the movement of the voltage platform along with aging. For the vehicle-mounted power battery pack, the voltage and the temperature can be conveniently and accurately measured, so that the characteristic parameters extracted based on the DTV curve have more representation capability on fault detection and state prediction. By using the DTV curve, a first peak value and a peak position, a second peak value and a peak position (the second peak is generally obtained, and the later peak is not suitable for being used as characteristic input due to overlarge fluctuation error of data) of the DTV curve can be extracted, and variables such as the area enclosed by the DTV curve and an abscissa axis are used as characteristic parameters.
The mathematical expression of the incremental capacity analysis ICA curve is as follows:
the ICA curve reflects capacity attenuation of the battery caused by aging or other reasons by using capacity change in the constant-current charging and discharging process, the change of voltage difference reflects the movement of a voltage platform along with aging, and a similar DVA curve is the reciprocal of an ICA curve expression. Similarly, the ICA curve can be used for extracting the first peak value and the peak position, the second peak value and the peak position, and the area and other variables surrounded by the abscissa axis are taken as the characteristic parameters.
The mathematical expression of the differential voltage analysis DV curve is as follows:
the voltage rises rapidly at the initial and final stages of battery charging; while in the middle of the charging curve the voltage rises slowly. In fact, the slowly rising plateau partially fills most of the capacity. The DV curve can convert a voltage plateau region where the voltage curve rises slowly and the reaction inside the battery is severe into a peak in the capacity increment curve that is easy to observe and analyze, and further, a small change that is not easy to observe on the voltage curve can be reflected on the DV curve. Similarly, the first peak value and the peak position, the second peak value and the peak position of the DV curve can be extracted by using the DV curve, and variables such as the area enclosed by the DV curve and the abscissa axis are used as characteristic parameters.
The charging and discharging voltage platform analysis UU curve is a scattered point image of a charging and discharging voltage platform point of each cycle of the battery on a two-dimensional coordinate, the abscissa and ordinate axes are the charging voltage platform point and the discharging voltage platform point respectively, and the numerical value is equal to
And
whether the battery has capacity diving phenomenon can be identified by observing the bending angle of the tail part of the UU graph curve, and whether the battery aging is caused by internal resistance increase or lithium precipitation can be detected by observing the integral swinging angle of the UU graph curve. The angle in the UU diagram can be used as the characteristic parameter of the input.
The changes of parameters such as ohmic resistance, polarization capacitance and the like obtained by parameter identification can be used as input characteristic parameters, and the changes of impedance and capacitance reactance values in Electrochemical Impedance Spectroscopy (EIS) detection can be simulated approximately, so that the phenomena such as electrode process dynamics, double electric layers, diffusion and the like are analyzed, and the mechanism changes such as reaction electrode materials, solid electrolytes, conductive polymers, corrosion protection and the like are analyzed.
The five types of electrochemical parameters (curves) can be obtained by calculating and fitting macroscopic parameters without configuring other equipment. The introduction of electrochemical parameters can effectively expand dimension for the input of a machine learning model and introduce more microscopic representations, the coupling of macroscopic parameters and microscopic parameters can enable characteristic parameters to contain more effective information, and the prediction accuracy of the model is fundamentally improved, wherein the characteristic parameters refer to parameters in a characteristic matrix, the characteristic matrix mentioned here is some characteristic quantities of time scale actually, time depends on the sampling frequency (10ms/1000ms and the like) of hardware, the characteristic quantities are extracted DTV peak position, ICA peak position and other N variables, and can be expressed mathematically: the matrix of sampling times N, here written as the feature matrix, is generally referred to as an input quantity as the feature matrix in consideration of the concepts of machine learning models, big data, and the like mentioned later.
The coupling of the macroscopic parameters and the microscopic parameters is a simpler coupling and is represented by the following three points:
the macroscopic parameters are input to the microscopic parameter processing module for calculation to obtain the required microscopic parameters.
And the macroscopic parameters still need to be used for judging faults of overvoltage, undervoltage, overcurrent, undercurrent, overtemperature, undertemperature, short circuit, open circuit and the like of the battery system, are coupled with the microscopic parameters in final early warning and diagnosis, and output an alarm signal.
When the battery system is over-temperature, the corresponding current-voltage curve also changes, so that the microscopic parameters also change except the change detected in the macroscopic temperature measurement; when the capacity of the battery system is attenuated, macroscopic parameters can be reflected in the capacity attenuation, the internal resistance is increased, and the voltage is reduced, which are difficult to detect, even if the battery is detected, the battery aging can not be proved powerfully, but the macroscopic parameters can be obviously reflected by the detection of the microscopic parameters, so that the macroscopic parameter coupling results can mutually compensate the short plates.
In the above embodiment, before the macro parameters and the micro parameters are uploaded to the cloud big data platform, edge calculation is performed at the output end of the vehicle-mounted power battery, which is mainly responsible for format verification of the macro parameters acquired by the sensor and the electrochemical micro parameters subjected to calculation processing, that is, a null value (NaN) of data that is not acquired due to a sensor failure is assigned, and data pre-cleaning can be performed by using methods such as a sliding window average value in a certain time interval or interpolation of a near point, for example, if no input or an obvious error input occurs at some sampling time because the sensor is sometimes interfered by environmental or human factors, if the null value or the error value is brought into the DTV, an incalculation result (shown as a glitch, a NaN null value, etc. on a curve) occurs in the ICA formula calculation, the method is actually a smoothing algorithm of filtering, and an interpolation method of a near point is that a non-null value which is close to the null value and can be nearest is directly assigned to the null value. The edge calculation is a ring of an end-edge-cloud framework, the data pre-cleaning work can be realized in the edge calculation device, the cloud data cleaning is performed through cloud calculation by a cloud large data platform, namely, the second-stage data cleaning work is completed at the cloud end, and the reliability of the data can be increased through two-stage data cleaning of the edge end and the cloud end.
In the above embodiment, the microscopic parameters after data pre-cleaning are the same as the macroscopic parameters, and real-time outlier detection is also performed on the cloud big data platform to realize rapid alarm on thermal runaway and other fault types, but the alarm is not required when the outlier occurs, the outlier alarm needs to meet the requirement that a plurality of characteristic variables (i.e., data streams which are real values) simultaneously or obviously appear in a certain time interval to trigger, if only the outlier appears in an individual characteristic variable, it is possible that a data acquisition or processing process has a fault, and then, through an error compensation strategy, individual data which are obviously wrong are cleaned or deleted, so that the logic of the data is correct, and the wrong data are prevented from being substituted into a machine learning model to cause errors.
Specifically, the cloud big data platform is provided with a machine learning model trained by a large amount of laboratory data and massive historical data, the model evolves synchronously with physical entities of the real world, and the model is trained and updated in real time according to data collected by the edge computing device. Specifically, the machine learning model is an LSTM neural network, the LSTM neural network has a better effect on battery capacity attenuation prediction, the LSTM neural network is a variant of an RNN (neural network) and is good at processing space-time sequence problems, the LSTM neural network mainly aims at solving the problems of gradient extinction and gradient explosion in a long sequence training process, compared with the common RNN, the LSTM neural network can have a better battery state prediction effect in a longer sequence, and on the basis of the LSTM neural network, in order to highlight the influence of certain input characteristics on a prediction result, a further evolution of the LSTM neural network can be used for driving the clamped-LSTM neural network, so that the model prediction accuracy can be effectively improved, and the invention is not particularly limited. After vehicle-end data are transmitted to a cloud big data platform, a machine learning model can train and analyze real-time data streams, the real-time fault alarm is mainly realized through outlier detection, the number of the outliers is compared with a preset threshold value of an alarm grade, whether the alarm and the alarm grade are judged according to a comparison result, the number of the outliers corresponds to the alarm grade one by one, and an alarm result is obtained through comparing the number of the outliers with the preset threshold value of the corresponding alarm grade in real time or within a certain time interval. The machine learning model can also predict the battery state in a future period of time by utilizing real-time data, such as the state of charge (SOC), the state of energy (SOE), the state of power (SOP), the service life prediction (SOH) and the like, and the high-dimensional input data containing a large amount of information can realize more accurate state prediction and provide guarantee for energy management and control of the whole vehicle.
In summary, the application process of the battery analysis method based on electrochemical parameters is as follows:
(1) macroscopic parameter information including current, voltage, temperature, capacity and internal resistance is acquired according to various sensors at a vehicle end, the macroscopic parameters are synchronously transmitted to an electrochemical microscopic parameter calculation device, corresponding real-time characteristic parameters are calculated by utilizing various curve mathematical expressions, and the real-time characteristic parameters are sent to a cloud large data platform through data pre-cleaning at the vehicle end, wherein the macroscopic parameters are acquired in real time, the microscopic parameters are calculated in real time, and the calculation requires time but can be considered to be real-time;
(2) data transmitted to the cloud big data platform are subjected to data cleaning with higher computation amount, the effectiveness of subsequently used data is guaranteed, and the data cleaning method includes but is not limited to spare assignment, error value correction and logic verification. The spare assignment is the method of the average value of the sliding window, the interpolation of the adjacent points and the like; error correction means that when the obtained data has obvious errors, the error is corrected by methods such as sliding window average value and near point interpolation, for example, the voltage of a battery is 3.6-4.3V, and the acquired voltage is less than 3.6V or more than 4.3V, the error is obviously corrected, and the error mainly appears in the acquisition errors of the sensor; in the battery management system, if a logical paradox occurs to a calculation closed loop of a detection circuit, nodes with errors need to be checked, for example, voltages of parallel batteries should be the same, but a certain branch is too large or too small, a logical error occurs at the moment, and resampling is needed for numerical value checking;
(3) the data cleaned by the data is subjected to fault real-time alarm based on an outlier detection algorithm in a cloud big data platform, and whether alarm is given or not and alarm level is determined by comparing whether the real-time data has outliers with multiple indexes in a short period or not;
(4) finally, the state of the battery is predicted by using a machine learning model of the cloud platform, such as the state of charge (SOC), the state of energy (SOE), the state of power (SOP), the service life prediction (SOH) and the like, so that a reliable basis is provided for the formulation of the energy management strategy of the whole vehicle, the fault information suspected of being outlier is compared with a fault set through a clustering method, the type of the fault which possibly occurs is judged, and a driver is reminded to make required countermeasures.
The present invention also provides a battery analysis system based on electrochemical parameters, which corresponds to the above battery analysis method based on electrochemical parameters, and can be regarded as a system for implementing the above battery analysis method based on electrochemical parameters, and the repeated parts are not repeated. As shown in fig. 2, the system includes: the macroscopic parameter acquisition device is used for acquiring macroscopic parameters of the battery; the microscopic parameter calculating device is used for calculating the microscopic parameters of the battery based on the electrochemical principle through a curve fitting algorithm according to the macroscopic parameters of the battery; the data cleaning device is used for pre-cleaning data of the battery macroscopic parameters and the battery microscopic parameters through edge calculation to obtain pre-cleaned data and sending the pre-cleaned data to the cloud big data platform, and performing cloud data cleaning on the pre-cleaned data sent to the cloud big data platform to obtain data streams which are real values; the outlier detection device is used for detecting and calculating the data streams which are all real values through an outlier detection algorithm to obtain the number of outlier data; the real-time alarm device is used for comparing the number of the cluster data with a preset threshold value of the alarm level and determining whether to alarm and the alarm level according to the comparison result; and the detection and prediction device is used for inputting the data streams which are all real values into the trained machine learning model, comparing the fault information of the outlier with the fault set through a clustering method, and predicting the fault type according to the clustering result so as to realize the state estimation of the battery.
Specifically, the macroscopic parameter acquisition device acquires the macroscopic parameters of the battery through a plurality of sensors at the vehicle end, and the macroscopic parameters comprise: current, voltage, temperature, capacity and internal resistance of the battery; the microscopic parameters calculated by the microscopic parameter calculating means include: the system comprises differential thermal voltage analysis curve parameters, incremental capacity analysis curve parameters, differential voltage analysis curve parameters, charging and discharging voltage platform analysis curve parameters and parameter identification electrochemical impedance spectrum parameters.
Based on the embodiment of the invention, the electrochemical parameter-based battery analysis system comprises a macroscopic parameter acquisition device, a microscopic parameter calculation device for calculating electrochemical principle-based battery microscopic parameters according to battery macroscopic parameters, a data pre-cleaning device for pre-cleaning the battery macroscopic parameters and the battery microscopic parameters to obtain pre-cleaned data, a cloud-side big data platform for cleaning the pre-cleaned data sent to the cloud-side big data platform with higher computation amount of cloud-side data, a data cleaning device for cleaning the two-stage data to obtain real-value data streams, an outlier detection device for detecting and calculating the real-value data streams through an outlier detection algorithm to obtain the number of outliers, an real-time alarm device for comparing the number of the outliers with a preset threshold value of an alarm grade, determining whether to alarm and the alarm grade according to the comparison result, and inputting the real-value data streams into the real-value data streams Comparing the fault information of the occurrence of the outlier with a fault set in a machine learning model by a clustering method, predicting the fault type which is possibly occurred according to the clustering result by a detection prediction device, mutually coordinating the devices of the system, coupling macroscopic parameters with microscopic parameters by a microscopic parameter calculation device, leading a data stream input into a cloud big data platform to contain more effective information, fundamentally improving the real-time prediction precision of the machine learning model, training and analyzing the real-time streaming data after the data stream is transmitted to the cloud big data platform, giving a fault alarm in real time by an outlier detection device, comparing the fault information of the occurrence of the outlier with the fault set by the machine learning model by the clustering method, predicting the fault type which is possibly occurred according to the clustering result, and predicting the battery state of a period of the future by the data stream input in real time, more accurate state prediction can be realized, and the energy management and control of the whole vehicle are guaranteed; the micro parameter calculation device calculates and obtains the battery micro parameters based on the electrochemical principle, and the electrochemical parameters can represent the fundamental characteristics of the internal reaction of the battery, so that the battery state can be more accurately and comprehensively depicted, the reliability is higher, compared with the traditional single macro parameter analysis and detection method, the method is more comprehensive, the more accurate prediction effect can be realized, the electrochemical performance, the thermodynamic performance, the mechanical fault and the like of the battery can be effectively disclosed, the more comprehensive real-time fault alarm effect is realized, the failure rate and the false alarm rate are reduced, the reliability of data can be increased through the data cleaning device, higher quality input is provided for a machine learning model, and the quality improvement of the battery performance detection and prediction result can be realized by utilizing the machine learning model by depending on a powerful cloud large data platform.
Specifically, the data cleaning device comprises an edge calculating device, is arranged at the output end of the vehicle-mounted power battery and is used for carrying out format verification on the collected battery macroscopic parameters and the electrochemical microscopic parameters subjected to calculation processing and assigning values to null values of data which are not collected due to sensor faults so as to realize data pre-cleaning.
Specifically, the characteristic parameters are obtained through a macroscopic parameter acquisition device and a microscopic parameter calculation device based on an electrochemical principle, data cleaning is carried out on the characteristic parameters through an edge calculation device and a cloud, the characteristic parameters are converted into data streams which are all real values and are transmitted to a cloud big data platform, the time points of a large number of simultaneously occurring outliers are marked as alarm identification points through an outlier detection device and fed back to a vehicle end for real-time alarm, the data streams enter a machine learning model through the cloud big data platform, accurate state prediction and fault type diagnosis and detection of future time can be achieved, and support is provided for optimizing whole vehicle energy management and early warning of faults.
The battery analysis system based on the electrochemical parameters is based on the electrochemical parameters, has characterization capability of coupling macroscopic characteristics and micro mechanisms, keeps a rolling optimization power system through a cloud big data platform and a built-in machine learning model, is constantly in a dynamic information receiving and self-updating optimization process, and can provide guarantee for safe and efficient operation of the battery.
It should be noted that the above-mentioned embodiments enable a person skilled in the art to more fully understand the invention, without restricting it in any way. Therefore, although the present invention has been described in detail with reference to the drawings and examples, it will be understood by those skilled in the art that various changes and modifications can be made therein without departing from the spirit and scope of the invention.
Claims (4)
1. A method for analyzing a battery based on electrochemical parameters, said method comprising the steps of:
the method comprises the following steps of obtaining battery macroscopic parameters through a plurality of sensors at a vehicle end, wherein the macroscopic parameters comprise: current, voltage, temperature, capacity and internal resistance of the battery;
calculating the microscopic parameters of the battery based on the electrochemical principle by a curve fitting algorithm according to the macroscopic parameters of the battery, wherein the microscopic parameters comprise: differential thermal voltage analysis curve parameters, incremental capacity analysis curve parameters, differential voltage analysis curve parameters, charging and discharging voltage platform analysis curve parameters and parameter identification electrochemical impedance spectrum parameters;
performing data pre-cleaning on the battery macroscopic parameters and the battery microscopic parameters through edge calculation to obtain pre-cleaned data, sending the pre-cleaned data to a cloud big data platform, and performing cloud data cleaning on the pre-cleaned data sent to the cloud big data platform to obtain data streams which are real values; the data pre-cleaning comprises the steps of carrying out format verification on collected battery macroscopic parameters and battery microscopic parameters subjected to calculation processing, and assigning a null value of data which are not collected due to sensor faults, and the cloud data cleaning comprises the steps of assigning the null value, correcting error values and carrying out logic verification;
detecting and calculating the data streams which are all real values through an outlier detection algorithm to obtain the number of outlier data;
comparing the number of the outlier data with a preset threshold value of an alarm level, and determining whether to alarm and the alarm level according to a comparison result;
and inputting the data streams which are all real values into a trained machine learning model, comparing the fault information of the outlier with a fault set by using the machine learning model through a clustering method, and predicting the fault type according to a clustering result to realize the state estimation of the battery.
2. The method according to claim 1, wherein the number of the outliers is compared with a preset threshold of an alarm level, and whether the alarm and the alarm level are given or not is determined according to the comparison result, specifically, the number of the outliers corresponds to the alarm level one by one, and the alarm result is obtained by comparing the number of the outliers with the preset threshold of the corresponding alarm level in real time or within a certain time interval.
3. The method according to claim 1 or 2, wherein the predicting the fault type according to the clustering result comprises predicting a state of charge (SOC), a state of energy (SOE), a state of power (SOP), and/or a state of life (SOH) according to the clustering result.
4. A battery analysis system based on electrochemical parameters, the system comprises a macroscopic parameter acquisition device for acquiring macroscopic parameters of a battery through a plurality of sensors at a vehicle end, and the macroscopic parameters comprise: current, voltage, temperature, capacity and internal resistance of the battery, characterized in that the system further comprises:
the microscopic parameter calculation device is used for calculating the microscopic parameters of the battery based on the electrochemical principle through a curve fitting algorithm according to the macroscopic parameters of the battery, and the microscopic parameters comprise: differential thermal voltage analysis curve parameters, incremental capacity analysis curve parameters, differential voltage analysis curve parameters, charging and discharging voltage platform analysis curve parameters and parameter identification electrochemical impedance spectrum parameters;
the data cleaning device is used for performing data pre-cleaning on the battery macroscopic parameters and the battery microscopic parameters through edge calculation to obtain pre-cleaned data and sending the pre-cleaned data to the cloud big data platform, and performing cloud data cleaning on the pre-cleaned data sent to the cloud big data platform to obtain data streams which are real values; the data cleaning device comprises an edge calculating device, a data pre-cleaning device and a data pre-cleaning device, wherein the edge calculating device is arranged at the output end of the vehicle-mounted power battery and is used for carrying out format verification on the acquired battery macro parameters and the battery micro parameters subjected to calculation processing and assigning values to null values of data which are not acquired due to sensor faults so as to realize data pre-cleaning; the cloud data cleaning comprises spare assignment, error value correction and logic verification;
the outlier detection device is used for detecting and calculating the data streams which are all real values through an outlier detection algorithm to obtain the number of outlier data;
the real-time alarm device is used for comparing the number of the cluster data with a preset threshold value of the alarm level and determining whether to alarm and the alarm level according to the comparison result;
and the detection and prediction device is used for inputting the data streams which are all real values into a trained machine learning model, comparing the fault information of the outlier with a fault set through a clustering method, and predicting the fault type according to a clustering result so as to realize the state estimation of the battery.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010314133.1A CN111624494B (en) | 2020-04-20 | 2020-04-20 | Battery analysis method and system based on electrochemical parameters |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010314133.1A CN111624494B (en) | 2020-04-20 | 2020-04-20 | Battery analysis method and system based on electrochemical parameters |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111624494A CN111624494A (en) | 2020-09-04 |
| CN111624494B true CN111624494B (en) | 2021-08-06 |
Family
ID=72269860
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010314133.1A Active CN111624494B (en) | 2020-04-20 | 2020-04-20 | Battery analysis method and system based on electrochemical parameters |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111624494B (en) |
Families Citing this family (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112327194B (en) * | 2020-10-21 | 2021-09-24 | 北京航空航天大学 | Lithium battery capacity diving identification method and device |
| CN112622676A (en) * | 2020-11-09 | 2021-04-09 | 特瓦特能源科技有限公司 | Monitoring method and system for power battery safe charging |
| CN112595986A (en) * | 2020-11-26 | 2021-04-02 | 同济大学 | Battery service life diagnosis method based on differential thermal voltammetry curve characteristics |
| CN112699113B (en) * | 2021-01-12 | 2022-08-05 | 上海交通大学 | Industrial manufacturing process operation monitoring system driven by time sequence data stream |
| KR20220114915A (en) * | 2021-02-09 | 2022-08-17 | 주식회사 엘지에너지솔루션 | Apparatus and method for diagnosing battery |
| CN113191677A (en) * | 2021-05-21 | 2021-07-30 | 中电工业互联网有限公司 | Vehicle-mounted storage battery alarming method capable of being dynamically configured |
| CN113839449A (en) * | 2021-10-29 | 2021-12-24 | 蜂巢能源科技有限公司 | Safety control method and control system for energy storage system |
| CN113884838A (en) * | 2021-11-04 | 2022-01-04 | 中汽数据有限公司 | Power battery insulation fault diagnosis method, equipment and storage medium |
| CN114325403B (en) * | 2021-12-03 | 2023-08-11 | 北京航空航天大学 | Lithium ion battery life detection method and system based on electrochemical impedance spectrum test |
| CN114325433A (en) * | 2021-12-03 | 2022-04-12 | 北京航空航天大学 | Lithium ion battery fault detection method and system based on electrochemical impedance spectrum test |
| CN114254766A (en) * | 2021-12-10 | 2022-03-29 | 智己汽车科技有限公司 | Method and system for jointly training machine learning model at two ends of vehicle cloud |
| CN114387332B (en) * | 2022-01-17 | 2022-11-08 | 江苏省特种设备安全监督检验研究院 | Pipeline thickness measuring method and device |
| CN114497770B (en) * | 2022-01-26 | 2022-09-09 | 上海玫克生储能科技有限公司 | Method, system and terminal for analyzing state of battery box in battery cluster |
| CN114460470A (en) * | 2022-01-26 | 2022-05-10 | 上海玫克生智能科技有限公司 | Battery pack state analysis method and system based on voltage and terminal |
| CN114460469A (en) * | 2022-01-26 | 2022-05-10 | 上海玫克生智能科技有限公司 | Battery state analysis method, system and terminal based on voltage and current |
| CN114734873B (en) * | 2022-04-18 | 2024-05-28 | 北京航空航天大学 | Power battery monomer thermal runaway early warning method based on cloud online data |
| WO2024007188A1 (en) * | 2022-07-06 | 2024-01-11 | 宁德时代新能源科技股份有限公司 | Battery parameter acquisition method and related apparatus |
| CN115149123B (en) * | 2022-07-28 | 2023-04-14 | 上海玫克生储能科技有限公司 | Lithium battery module consistency analysis method and system and storage medium |
| CN114994543A (en) * | 2022-08-01 | 2022-09-02 | 湖南华大电工高科技有限公司 | Energy storage power station battery fault diagnosis method and device and storage medium |
| CN115420401A (en) * | 2022-08-25 | 2022-12-02 | 上海玫克生储能科技有限公司 | Early warning method, early warning system, storage medium and electronic equipment |
| CN117092526B (en) * | 2023-10-20 | 2023-12-15 | 广东采日能源科技有限公司 | Training method and device of battery fault early warning model and electronic equipment |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105807226A (en) * | 2014-12-31 | 2016-07-27 | 北京航天测控技术有限公司 | Lithium ion battery SOC prediction method and device based on equivalent circuit model |
| CN106556802A (en) * | 2016-11-01 | 2017-04-05 | 东软集团股份有限公司 | A kind of accumulator battery exception cell recognition methodss and device |
| CN109116244A (en) * | 2018-06-29 | 2019-01-01 | 深圳市科列技术股份有限公司 | A kind of judgment method and device of power battery aging tendency |
| CN109738802A (en) * | 2018-12-06 | 2019-05-10 | 中车工业研究院有限公司 | A kind of the joint supervisory systems and method of vehicle-mounted energy-storage system |
| CN110058165A (en) * | 2017-10-17 | 2019-07-26 | 小利兰·斯坦福大学托管委员会 | The data-driven model of capacity of lithium ion battery decaying and life prediction |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10209314B2 (en) * | 2016-11-21 | 2019-02-19 | Battelle Energy Alliance, Llc | Systems and methods for estimation and prediction of battery health and performance |
-
2020
- 2020-04-20 CN CN202010314133.1A patent/CN111624494B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105807226A (en) * | 2014-12-31 | 2016-07-27 | 北京航天测控技术有限公司 | Lithium ion battery SOC prediction method and device based on equivalent circuit model |
| CN106556802A (en) * | 2016-11-01 | 2017-04-05 | 东软集团股份有限公司 | A kind of accumulator battery exception cell recognition methodss and device |
| CN110058165A (en) * | 2017-10-17 | 2019-07-26 | 小利兰·斯坦福大学托管委员会 | The data-driven model of capacity of lithium ion battery decaying and life prediction |
| CN109116244A (en) * | 2018-06-29 | 2019-01-01 | 深圳市科列技术股份有限公司 | A kind of judgment method and device of power battery aging tendency |
| CN109738802A (en) * | 2018-12-06 | 2019-05-10 | 中车工业研究院有限公司 | A kind of the joint supervisory systems and method of vehicle-mounted energy-storage system |
Non-Patent Citations (1)
| Title |
|---|
| Battery module capacity fade model based on cell voltage inconsistency and probability distribution;Yulong Zhang等;《Advances in Mechanical Engineering》;20170928;第9卷(第9期);1-13 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111624494A (en) | 2020-09-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111624494B (en) | Battery analysis method and system based on electrochemical parameters | |
| CN112241608B (en) | Lithium battery life prediction method based on LSTM network and transfer learning | |
| WO2021185308A1 (en) | Online determination method and system for state of health of power battery pack of electric vehicle | |
| Guo et al. | Online estimation of SOH for lithium-ion battery based on SSA-Elman neural network | |
| Song et al. | Remaining life prediction of lithium-ion batteries based on health management: A review | |
| EP4198531A1 (en) | Battery cell defectiveness and fire early diagnosis method and apparatus based on neural network | |
| CN103576096A (en) | Real-time assessment method and device for residual capacity of power battery of electric automobile | |
| CN107436409B (en) | Intelligent SOC prediction device for power battery of electric automobile | |
| CN113447828B (en) | Lithium battery temperature estimation method and system based on Bayesian neural network | |
| CN116609676B (en) | Method and system for monitoring state of hybrid energy storage battery based on big data processing | |
| Zhao et al. | Multi-step ahead voltage prediction and voltage fault diagnosis based on gated recurrent unit neural network and incremental training | |
| Wang et al. | Modeling and state of charge estimation of inconsistent parallel lithium-ion battery module | |
| Zhang et al. | Fault diagnosis of real-scenario battery systems based on modified entropy algorithms in electric vehicles | |
| CN114579659A (en) | System and method for estimating and sorting utilization potential of power battery by gradient utilization | |
| Singh et al. | State of charge estimation techniques of Li-ion battery of electric vehicles | |
| CN115799673A (en) | Energy storage battery management system with multidimensional sensing | |
| Vatani et al. | Cycling lifetime prediction model for lithium-ion batteries based on artificial neural networks | |
| CN115219907A (en) | Lithium battery SOC estimation method, system, medium, equipment and terminal | |
| Yang et al. | State of Health Estimation for Power Battery Based on Support Vector Regression and Particle Swarm Optimization Method | |
| Saleem et al. | Investigation of Deep Learning Based Techniques for Prognostic and Health Management of Lithium-Ion Battery | |
| Yan et al. | SOC prediction method of wireless sensor nodes batteries based on Attention-LSTM | |
| An | New energy vehicle lithium battery life prediction method based on improved deep learning | |
| Sun et al. | Internal short circuit fault diagnosis for the lithium-ion batteries with unknown parameters based on transfer learning optimized residual network by multi-label data processing | |
| Cai et al. | DIICAN: Dual Time-scale State-Coupled Co-estimation of SOC, SOH and RUL for Lithium-Ion Batteries | |
| CN117394409B (en) | Intelligent assessment method and system for equipment state of energy storage power station |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |